講演情報
[6D-05]Edge Classification on Imbalanced Multi-relational Graphs
*GONG ZHAOJIE1、DUAN YIJUN1、馬 強1 (1. 京都工芸繊維大学社会知能情報学研究室)
発表者区分：学生
論文種別：ロングペーパー
インタラクティブ発表：あり
論文種別：ロングペーパー
インタラクティブ発表：あり
キーワード：
graphdata、graph mining、imbalanced learning、edge prediction
On real-world multi-relational graphs, edge labels often follow an imbalanced long-tail distribution, posing significant challenges to edge representation learning and subsequent applications, such as edge classification and predicion. This arises because edge featurizers tend to be biased toward majority edge features, resulting in inadequate learning of features for minority class edges. To address this issue, we propose a novel method for tackling the edge representation learning on imbalanced multi-relational graphs. The key technical contributions include a synthetic minority edge generation approach inspired by SMOTE, combined with the construction of connecting links to enhance message passing on the augmented graphs. Extensive experiments on diverse real-world datasets demonstrate the effectiveness of the proposed method compared to competitive baselines.